RNA Velocity

RNA velocity is a computational framework that allows inference of the future state of individual cells in single-cell RNA-seq experiments. While RNA-seq data provides a static snapshot of RNA transcription, it reveals little about dynamic processes that occur during development or in response to stimuli. Addressing this challenge, La Manno and collaborators (2018) introduced RNA velocity, leveraging the observation that nascent (unspliced) RNA exhibits different dynamics compared to mature (spliced) RNA. Based on the analysis of the ratio of spliced and unspliced RNA, the direction of a cell state transition is predicted in the high-dimensional gene expression space, offering insights into cellular trajectories.

At its core, RNA velocity models transcription activity using splicing kinetics equations, assuming steady state or dynamical models. A simpler analogy is imagining a UMAP plot as a dynamic movie where cells move from one state to another over time. In this context, RNA velocity predicts those transitions. However, RNA velocity does not measure transcription directly but instead leverages the comparison of unspliced and spliced RNA to make inferences.

Unspliced molecules represent RNA that is being synthesized; the higher this proportion, the more active the transcription of particular genes. The steady-state RNA (spliced) represents the pool of RNA already synthesized. Based on the first, the magnitude of the second can be calculated, and transitions can be predicted accordingly. A useful analogy is motion in physics: the steady-state levels of RNA correspond to a cell’s current position, while nascent (unspliced) RNA reflects its acceleration. With these two components, RNA velocity calculates a cell’s future position (state) at time t + 1.

This method has multiple applications in diverse research areas, including:

• Developmental biology: Stem-cell transitions into diverse cell types.
• Cancer biology: Identifying cell types associated with more aggressive cancer.
• Cellular reprogramming: Studying pluripotency or direct cell-type conversions.
• Immune system dynamics: Unveiling immune cell dynamics in response to stimulation.
• Gene regulatory networks: Identification of regulators of cell state transitions.
• Disease progression and therapy: Unraveling cellular dynamics during therapy.

Computational Pipeline

Quantification

Data used in this tutorial corresponds to PBMCs from a healthy individual, described here.

1. Create a new mamba environment with the required software:

mamba create -n RNAvel python=3.11 numpy=1.23 pandas=1.5 scanpy=1.9 pyroe salmon alevin-fry
mamba activate RNAvel

    

2. Download genome and GTF files:

# Genome
wget https://ftp.ensembl.org/pub/current_fasta/homo_sapiens/dna/Homo_sapiens.GRCh38.dna.toplevel.fa.gz 

# GTF
wget https://ftp.ensembl.org/pub/current_gtf/homo_sapiens/Homo_sapiens.GRCh38.113.chr_patch_hapl_scaff.gtf.gz

    

3. Create a spliced reference genome:

pyroe make-splici "$GENOME" "$GTF" 151 hs_GRCh38_113_splici_python --flank-trim-length 5 --filename-prefix splici

    

The above command will produce a folder named hs_GRCh38_113_splici_python, containing files like:

• clean_gtf.gtf
• gene_id_to_name.tsv
• splici_fl146.fa
• splici_fl146_t2g_3col.tsv

References

1. La Manno, G., Soldatov, R., Zeisel, A., et al. RNA velocity of single cells. Nature, 560, 494–498 (2018). DOI
2. Domínguez Conde C, Xu C, Jarvis LB., et al. Cross-tissue immune cell analysis reveals tissue-specific features in humans. Science. 2022. DOI: 10.1126/science.abl5197.
3. Xu C, Prete M, Webb S., et al. Automatic cell-type harmonization and integration across Human Cell Atlas datasets. Cell, 2023. DOI: 10.1016/j.cell.2023.11.026.